Social integration and mental health of Somali refugees in the Netherlands: the role of perceived discrimination

Social integration refers to “how migrants forge social relations which enhance their connectedness with the place in which they settle and the wider society around them” [11]. It is a multi-faceted concept [12], with key facets being proficiency in the language of the destination country and informal contacts with destination country natives without a migration background [13, 14]. As argued by Dagevos [12], social integration can be seen as related to, yet distinct from, cultural integration, i.e., the extent to which immigrants ascribe to the host society’s prevailing moral standards and values, and the extent to which they identify as members of the host society [14, 15]. Although the central focus of the current study is not on cultural integration, we will take aspects of cultural integration – specifically ethnic identification – into account when assessing the associations between facets of social integration and refugees’ mental health. We do so because social integration and cultural integration are closely intertwined, and because prior research suggests that the latter is associated with refugees’ mental health [7].

Social integration may benefit mental health in multiple ways. When social networks are diverse and include members of the native population of the destination country in addition to co-ethnics, this may open ways to important resources, e.g., information, socially distant from the refugee that could benefit their life chances, and potentially their health, in the destination country [16, 17]. In contrast, inadequate local social networks may constitute a barrier for refugees to explore services and opportunities potentially available to them [18].

Poor command of the language in the destination country hinders developing informal contacts and social relationships with destination country natives, and forms a direct barrier to access to community services [19‐21]. Low language proficiency has also been linked to raised feelings of loneliness among Somali immigrants in Western countries [18, 22], and loneliness is, in turn, a known risk factor for depressive symptoms [23]. Moreover, communication difficulties arising from low language proficiency may complicate the contact with a mental healthcare professional, which may result in suboptimal quality of the care received [24]. This may be stressful and could, depending on the individual’s coping behavior, result in negative emotions, which, ultimately, is detrimental for mental health. Refugees may also refrain from seeking mental healthcare entirely due to fear that they cannot communicate adequately with healthcare professionals, or because they are embarrassed about their language abilities [25‐27]. Low language proficiency may thus result in inadequate – or even absent – treatment of emerging mental health issues, which, in turn, can be expected to translate in suboptimal mental health [28].

The premise of the current study is that perceived discrimination may partly shape the impact of social integration on mental health among Somali refugees. Somali refugees are a group with multiple minority statuses, and they may therefore be target of discrimination on multiple grounds (e.g., race/ethnicity or Islamic beliefs) [29]. The negative association between experiences of discrimination and mental health is well-established [30‐34], also in Somali refugees [35‐38]. Perceptions that society is discriminatory and everyday experiences of minor acts of unjust treatment because of one’s social identity are chronic stressors. More apparent personal discriminatory acts are moreover acute stressors that are superimposed on this chronic stress, and the combined stress resulting from discrimination is detrimental for mental health [39].

Experiences of discrimination may either be a catalyst or a nullifier of potential beneficial mental health effects of social integration. Assimilation theory [40] would lead one to expect that discrimination plays the former role. Social integration can reduce the social distance between ethnic minority and native majority members [41]. Informal contact and familiarity with ethnic minority members who have proficiency in the language of the destination country and orientate towards the destination society may make ethnic majority group members more accepting towards minority members. As a consequence, attenuation of prejudice and discrimination may be expected [17, 40, 42]. Moreover, socially integrated ethnic minority members might feel more recognized and develop positive attitudes towards the native majority group. Assimilation theory would thus lead one to hypothesize that the beneficial mental health effects of destination country language proficiency and informal contacts with destination country natives can partly be attributed to reduced experiences of discrimination.

In stark contrast with what assimilation theory would lead one to expect, a phenomenon coined the integration paradox has recently been noted by scholars focusing on the links between social integration and discrimination [43, 44]. The integration paradox postulates a positive, rather than negative, association between migrants’ aptitude to integrate and experiences of discrimination [43, 44], and it posits that the greater exposure to discrimination among more strongly integrated migrants may result in disengagement in and resentment towards the destination society [45].

There are various explanations for a positive association between social integration and perceived discrimination. Integrated ethnic minority members are more exposed to contact with native majority members in the public context than their counterparts who mainly surround themselves with members from their own ethnic group. This may increase the sense of intergroup competition and increase the desire to strongly support in-group interests. Moreover, discrimination is more easily detected by those who speak the language of the destination country well [43]. Integrated ethnic minorities in the Netherlands are also to a larger extent confronted with the negative public view towards and political debates about ethnic minorities [46], which may increase feelings of discrimination [43]. Warfa et al. [29] noted frustration among many Somali refugees with professional skills and a very good command of the English language in the United Kingdom and the United States about the lack of recognition for their skills in the destination countries. This frustration, the authors argued, put the refugees at increased risk of resentment and psychological distress [29]. Following this reasoning, it may be hypothesized that potential beneficial mental health effects of destination country language proficiency and informal contacts with destination country natives are, in part, cancelled out by stronger experiences of discrimination among those with better destination country language proficiency and more informal contacts with natives.

The current study draws on data from the controlled circulation file of the 2015 edition of the Survey Integration Minorities (SIM) (Dutch: Survey Intergratie Minderheden) [47, 48]. SIM is a repeated cross-sectional survey commissioned by the Dutch Ministry of Social Affairs and Employment and conducted by the Netherlands Institute for Social Research. It focuses on the structural and socio-cultural position of migrants in the Netherlands. The third round of SIM (SIM2015), collected between January and June 2015, was the first to include a subsample of migrants of Somali origin. Other groups interviewed were people with Turkish, Moroccan, Surinamese, Antillean, Polish and native Dutch backgrounds.

Andriessen and Kappelhof describe in detail how the Somali subsample was recruited [47]. Statistics Netherlands randomly selected 1747 individuals of Somali origin (i.e., born in Somalia or at least one parent born in Somalia) from the Dutch population registers. Of these persons, 10 could not be contacted, either because they were deceased or because they no longer had a valid address. The 1737 remaining persons were sent a survey invitation, and 626 of them participated in the survey (response rate: 36.0%). Of these respondents, 234 filled in the survey online and 392 participated in computer-assisted personal interviews (CAPI). The questionnaires for the face-to-face interviews and the web survey were identical, and both were available in Dutch as well as in Somali. Face-to-face data collection was done by 15 interviewers with a Somali background.

Respondents were informed about the goals of the research, and it was emphasized that information would be anonymized and that participation in the survey and consent were voluntary. In the controlled circulation file of the dataset, selected information (e.g., age) was recategorized in broader categories in order to render identification of individual respondents impossible. The Research Ethics Review Committee of the Erasmus School of Health Policy & Management at Erasmus University Rotterdam assessed and approved the secondary analysis of the controlled circulation file of the SIM2015 dataset on which we report here (Reference: 21-033).

Given our focus on refugees, we excluded second-generation migrants (n = 51) and migrants who migrated for other reasons than war, political reasons or religious persecution (n = 158). Given that information on formal refugee status was not available, we had to rely on self-reported migration reasons as a proxy measure. The exclusion procedure described here resulted in a reduction of the analytical sample to 417 respondents. Supplied analytical weights were used to account for selective non-participation in the survey.

Mental health was measured with an alternative version of the Mental Component Summary of the 12-item Short Format Health Survey (SF-12) [49]. Respondents were asked how often in the last four weeks they “had a lot of energy”, “had difficulty to meet family or friends or plan other social activities due to physical health or emotional problems”, “felt calm and peaceful”, “felt downhearted and blue”, “participated less in activities or accomplished less than you would like as a result of any emotional problems”, and “were unable to do work or other regular daily activities as carefully as usual due to any emotional problems”. Response categories differed somewhat from the original SF-12: the four first-mentioned questions had four response categories ranging from (1) “all the time” to (4) “never”, and the final two questions had two response categories: (1) “yes” and (2) “no”. Where necessary, answers were recoded, so that higher values corresponded with better mental health. Following the procedure described by Schellingerhout [50], a summary scale ranging from 0 to 100 was subsequently derived.

Two facets of social integration were considered: Dutch language proficiency and the frequency of informal contact with Dutch natives. Three questions assessed whether the respondents experienced difficulties when having a conversation in Dutch, when writing in Dutch, and when reading Dutch newspapers, letters or brochures. The response categories for each of the three questions were (1) “often”, (2) “sometimes” and (3) “never”. For the first question, “not able to speak Dutch” (4) was presented as an additional response category. Responses were summed into a consistent scale (Cronbach’s α = .80) ranging from 3 to 10, with higher scores indicating better Dutch language proficiency.

With regard to informal contact with Dutch natives, respondents were asked how often native Dutch friends or neighbors visited them, how often they spent leisure time with Dutch natives, and how often they had contact with native Dutch friends or acquaintances. For the first two questions, response categories ranged from (0) “never” to (2) “often”. Response categories for the third question ranged from (0) “Once per month or less” to (2) “daily”. Summing the answers on the three questions to resulted in an internally consistent scale (Cronbach’s α = .73) ranging from 0 to 6, with higher scores indicating more informal contact with Dutch natives. Perceived personal discrimination was captured with the question: “Have you ever been discriminated by the Dutch population? If so, how often did this happen?”. Response categories ranged from “never” (1) to “very often” (5).

In the multivariate analyses we controlled for gender, age, partner status, parenthood, educational attainment, employment status, religiosity and identification with Somalia and the Netherlands. These variables were considered potential confounders, because they may be expected to be associated with social integration as well as with mental health [6, 7, 32, 37, 44, 51‐54].

Four age categories were distinguished: 15-24 years old, 25-34 years old, 35-44 years old, and 45 years and older. With regard to educational attainment, we distinguished three categories: (1) primary education or less, (2) lower secondary education, (3) higher secondary or tertiary education. Partner status was measured with a dichotomous variable distinguishing between respondents with a partner living in the Netherlands and their counterparts without a partner or with a partner living in a different country than the Netherlands. A dichotomous variable was included to distinguish respondents with at least one child from their childless counterparts. Another dichotomous variable indicated whether or not respondents were in paid employment. Length of stay was calculated by deducting the year in which the respondent first migrated to the Netherlands from the interview year. Given the positively skewed distribution of length of stay in years, log transformation was subsequently performed. Two dichotomous measures were included to capture respondents’ identification with Somalia and the Netherlands. Both variables were coded as 1 for respondents who reported identifying themselves “strongly” or “very strongly” as Somali or Dutch, respectively, and as 0 if they reported identifying themselves as such only “somewhat”, “not”, or “not at all”. A final dichotomous variable was included to capture religiosity. Note that respondents in our sample almost exclusively (96.4%) reported being Muslim. We distinguished between respondents who attended religious services at least weekly and their counterparts who attended religious services less often. Sample characteristics are presented in Table 1.

Information on one or more variables of interest was missing for 88 respondents (22.1%) in the analytical sample. Information on mental health (n = 48) and informal contact with natives (n = 32) was missing most often. Multiple imputation using chained equations was used to deal with missing data [55]. The underlying assumption is that information was missing at random (MAR), i.e. that differences between the distributions of missing values and the distributions of observed values could be explained by variables included in the imputation model [56]. The findings from the substantive analyses on 20 imputed data sets were combined into a single set of results following Rubin’s rules [57], which take the variability in results between the imputed datasets into account. An overview of the sample characteristics after multiple imputation and weighting is provided in Additional file 1: Appendix A.

We estimated a set of ordinary least squares regression models to predict mental health. We first regressed mental health on Dutch language proficiency, informal contacts with natives, and the range of control variables described above. In a second model we added perceived discrimination as an explanatory variable. We performed formal mediation analyses to assess whether the effects of Dutch language proficiency and informal contacts with natives on mental health were mediated (or suppressed) by perceived discrimination. Following the procedure proposed by Preacher and Hayes [58], bootstrapping was used to estimate the indirect effects of Dutch language proficiency and informal contacts with natives on mental health via perceived discrimination.